基于模型的三维物体识别

实现了一个完整的基于模型的三维物体识别系统,它可识别灰度图象中包含的物 体,如对遮挡加以限制,还可识别被遮挡的物体.该系统能实现物体的自动建模,也可先定性 识别某一物体的立体图对以获取高层知识,然后在高层知识的指导下准确地匹配立体图对中 相对应的特征.此外,还提出了利用最能表示物体特征的表面(特征面)来识别物体的方法,以 提高系统抗噪声的能力.大量实验证明,该系统具有相当的稳健性.     

由深度数据重建三维物体的一种方法

讨论由深度数据重建三维物体的一种新方法.其基本思想是根据由基于物体旋转的主动 式三角法三维信息获取系统所获取的物体深度数据作为该物体的型值点.再根据型值点进行 网格划分,以便形成满足一定约束条件的空间拓扑网格.最后使用有理Bezier三角曲面片与 矩形曲面片相结合的方法进行曲面拟合与逼近.     

基于遗传算法与模型投影的三维物体匹配

基于遗传算法与模型投影的三维物体匹配,实际上是三维参考模型投影参数的搜索和优化过程.本文采用粗细层次结合的策略,即先基于模型外轮廓,再进行整体边缘匹配的方法,使模型参数在优化过程中迅速得到正确的初始值,提高遗传算法优化的效率.相似度评价函数采用34Chamfer距离变换,提出使用距离变换的标记图方法建立目标图像边缘点与其距离图的对应关系,降低噪点的影响,目标图像的特征也可以通过对应关系在匹配过程中被引用.实验结果显示本文方法在三维物体匹配中的有效性.     

一种计算空间封闭曲面正侧法向量的方法

在三维物体表面光照处理,三维曲面上曲线所围区域的计算以及三维物体体积计算等应用中,需要计算曲面正侧法向量,针对三角形网格表示的空间封闭曲面,本文设计了一种计算曲面正侧法的向量的简明算法。     

基于相位恢复的三维可视化应用研究

空间曲面的三维数字化测量在工业检测、虚拟现实、影视艺术以及文物保护等领域具有重要的应用价值。提出一种实现自由曲面三维数字化的方法,该方法不仅能重建物体的几何模型,而且能实现物体表面色彩等纹理属性的恢复。利用结构光照明重建空间三维点云数据,并利用匹配算法建立物体的几何模型;然后通过交互式选取点优化策略进行纹理映射的方法实现目标物体的真实感三维重建。与传统的方法相比,该方法能够精确地实现目标物体的纹理重建,避免了特征检测困难或粘贴标准点等方法的数据处理困难以及后期数据的人工等操作,实验结果也验证了该方法的有效性。     

基于SIFT和Hu特征融合的单目视觉识别算法研究

研究了机器视觉技术在三维物体识别定位问题中的应用。利用Visual C＋＋编程软件建立了摄像机标定界面,实现对摄像机进行快速标定的功能。提出了一种SIFT特征和Hu不变矩融合算法,该算法是一种融合了局部特征和全局特征的算法。其中全局特征反映了对三维物体图像的整体信息进行粗略匹配和定位,局部特征可以在全局特征中进行更准确特征匹配,该算法对伸缩、旋转和平移等有很好的抵抗能力。实验结果表明,该视觉算法可以有效解决三维物体匹配问题,并且有效提高了系统的识别速率和效率,满足物体识别的目的。     

三维物体曲面的识别研究

本文对三维物体的特征进行了分析，在此基础上，提出了一种能够识别曲面形状的算法，有效地识别出锥面、柱面、平面等曲面，并给出了程序框图。     

带有未知匹配的图像线特征三维重建方法的研究*

为了解决传统三维重建过程中线特征匹配速度慢、建模与匹配过程脱节的问题,提出了基于相关系数的图像线特征匹配的方法,运用此方法对同一物体的三幅图像进行了特征匹配和三维重建。建立了目标物体的三维模型,达到了对物体线特征匹配与三维重建同步进行的目的。通过对所得实验数据分析,可以证明此算法对于物体的三维重建是有效的。     

基于形状拆分的机械零件三维模型检索

提出一种基于形状拆分的机械零件三维模型检索方法。首先根据零件名称和功能,将机械零件模型库按传统方式划分为四大类,即轴套类、盘盖类、叉架类和箱体类,三维模型之间的匹配在同一类模型中进行;随后进行三维模型外轮廓扫描,将整个三维模型拆分成若干平面体、回转体和曲面体,根据同一型体之间特征数量和参数的不同,进行单一型体之间的匹配;最后根据单一型体的数量和匹配对的相似度,综合计算整个模型的相似度。单一型体的定义和整个匹配算法均是围绕着形状和工艺相似进行的。通过实例验证了该方法的有效性。     

自由曲面视觉测量标志点三维匹配方法研究

提出一种适用于自由曲面视觉测量的标志点三维匹配方法.该方法首先通过编码标志点确定局部测量区域在物体上的初始位置,获得可能匹配点集;然后根据非编码标志点与编码标志点之间的距离、角度、方向等特征,获得全局坐标系与局部坐标系之间的初始转换关系;最后根据此初始转换关系求解最终匹配点集,完成标志点的三维匹配.该方法适用于基于标志...     

具有相切面的曲面立体线画图标记研究

为了更好地理解三维景物，介绍了前人对线画图标记的研究成果，同时建立了一种标记具有相切面的曲面立体线画图的方法，并给出了具有相切面的曲面立体线画图的标记规则。由于线画图中的节点是由三面角构成的顶点的投影，因此对于画出隐藏线的具有相切面的曲面立体线画图，其合法的节点标记形式有38种，其中Y型节点有6种，W型节点有18种，V型节点有14种。新的标记方法适用于由平面立体和曲面立体组合而成的复杂立体线画图。     

3D reconstruction of complex curved objects from line drawings

An active research topic in computer vision and graphics is developing algorithms that can reconstruct the 3D surface of curved objects from line drawings. There are a number of algorithms have been dedicated to solve this problem, but they can't solve this problem when the geometric structure of a curved object becomes complex. This paper proposes a novel approach to reconstructing a complex curved 3D object from single 2D line drawings. Our approach has three steps:(1) decomposing a complex line drawing into several simpler line drawings and transforming them into polyhedron;(2) reconstructing the 3D wireframe of curved object from these simpler line drawings and generating the curved faces;(3) combining the 3D objects into the complete objects. A number of examples are given to demonstrate the ability of our approach to successfully perform reconstruction of curved objects which are more complex than previous methods.     

Exploiting parallelism in 3-D object recognition using the connection machine

This paper addresses the problem of parallelism in 3-D object recognition and localization using a fine-grained SIMD architecture such as the Connection Machine^TM. The input images of partially occluding three-dimensional objects. The objects are made up of piecewise compositions of curved surfaces. The surfaces of interest are cylindrical, conical and spherical surfaces since they cover a large portion of objects encountered in an industrial environment. Qualitative classification of surfaces based on the signs of the Mean and Gaussian curvature is used to come up with Dihedral feature junctions. Dihedral feature junctions are shown to be robust to occlusion and offer a viewpoint independent scheme for modeling the object models and are well suited for matching and pose determination. Hough clustering is chosen as the constraint propagation mechanism on account of the ease of parallelization. Issues regarding parallelism on the Connection Machine^TM for every stage of the recognition and localization process such as segmentations, feature extraction matching and pose determination are discussed. Experimental results on the Connection Machine^TM bring out the advantages of exploiting parallelism for 3-D object recognition and localization on a fine-grained SIMD architecture.     

Interpreting line drawings of curved objects

In this paper, we study the problem of interpreting line drawings of scenes composed of opaque regular solid objects bounded by piecewise smooth surfaces with no markings or texture on them. It is assumed that the line drawing has been formed by orthographic projection of such a scene under general viewpoint, that the line drawing is error free, and that there are no lines due to shadows or specularities. Our definition implicitly excludes laminae, wires, and the apices of cones.A major component of the interpretation of line drawings is line labelling. By line labelling we mean (a) classification of each image curve as corresponding to either a depth or orientation discontinuity in the scene, and (b) further subclassification of each kind of discontinuity. For a depth discontinuity we determine whether it is a limb—a locus of points on the surface where the line of sight is tangent to the surface—or an occluding edge—a tangent plane discontinuity of the surface. For an orientation discontinuity, we determine whether it corresponds to a convex or concave edge. This paper presents the first mathematically rigorous scheme for labelling line drawings of the class of scenes described. Previous schemes for labelling line drawings of scenes containing curved objects were heuristic, incomplete, and lacked proper mathematical justification.By analyzing the projection of the neighborhoods of different kinds of points on a piecewise smooth surface, we are able to catalog all local labelling possibilities for the different types of junctions in a line drawing. An algorithm is developed which utilizes this catalog to determine all legal labellings of the line drawing. A local minimum complexity rule—at each vertex select those labellings which correspond to the minimum number of faces meeting at the vertex—is used in order to prune highly counter-intuitive interpretations. The labelling scheme was implemented and tested on a number of line drawings. The labellings obtained are few and by and large in accordance with human interpretations.     

The Interpretation of Line Drawings with Contrast Failure and Shadows

In line drawings derived from real images, lines may be missing due to contrast failure and objects with curved surfaces may cast shadows from multiple light sources.This paper shows that it is the presence of shadows, rather than contrast failure, that renders the line drawing labelling problem NP-complete. However, shadows are a valuable visual cue, since their presence is formally shown to reduce the average ambiguity of drawings. This is especially true when constraints concerning shadow formation are employed to differentiate shadow and non-shadow lines.The extended junction constraint, concerning straight lines colinear with junctions, compensates the loss of information caused by contrast failure. In fact, we observe the contrast failure paradox: a drawing is sometimes less ambiguous when lines are partly missing due to contrast failure.It is known that the coplanarity of sets of object vertices can be deduced from the presence of straight lines in the drawing. This paper shows that these coplanarity constraints are robust to the presence of contrast failure.     

曲面立体线图的补线技术研究

文中根据已有的曲面立体自然线图的标记原理，得到了曲面立体画隐线图的标记原理和方法。曲面立体的不完整线图中的L型节点是曲面立体完整线图中W型节点的退化。基于曲面立体画出隐藏线的完整线图标记原理和方法，详细分析并建立了曲面立体不完整线图中L型节点与曲面立体完整线图中W型节点的对应关系。提出了曲面立体不完整线图的标记和补线方法，结合算例详细论述了曲面立体不完整线图的标记和补线过程，从曲面立体不完整线图获得了合理的曲面立体完整线图。     

Stereo‐Consistent Contours in Object Space

Notebook scribbles, art or technical illustrations—line drawings are a simplistic method to visually communicate information. Automated line drawings often originate from virtual 3D models, but one cannot trivially experience their three‐dimensionality. This paper introduces a novel concept to produce stereo‐consistent line drawings of virtual 3D objects. Some contour lines do not only depend on an objects geometry, but also on the position of the observer. To accomplish consistency between multiple view positions, our approach exploits geometrical characteristics of 3D surfaces in object space. Established techniques for stereo‐consistent line drawings operate on rendered pixel images. In contrast, our pipeline operates in object space using vector geometry, which yields many advantages: The position of the final viewpoint(s) is flexible within a certain window even after the contour generation, e.g. a stereoscopic image pair is only one possible application. Such windows can be concatenated to simulate contours observed from an arbitrary camera path. Various types of popular contour generators can be handled equivalently, occlusions are natively supported and stylization based on geometry characteristics is also easily possible.     

三维物体识别及姿态测定的推理系统

本文介绍了三维物体识别及姿态测定的一种新技术，从物体空间域模型出发，通过约束推理及几何推理，在物体三维信息部分给定的条件下，推断预测图象模型，并通过实测的图象数据反馈，推断出隐含在图象中未给定的三维信息，最终实现三维物体识别及姿态测定。整个系统在ＶＩＣＯＭ机上用Ｃ语言完成。     

Frontal geometry from sketches of engineering objects: is line labelling necessary?

A tool which can quickly interpret line drawings (with hidden lines removed) of engineering objects as boundary representation CAD models would be of significant benefit in the process of engineering design. Inflation of the drawing to produce a frontal geometry, a geometric realisation of that part of the object visible in the drawing, is an important stage of this process.Previous methods of producing frontal geometries have relied on the technique of line labelling (labelling edges as convex, concave or occluding). Although restricted subsets of the line-labelling problem have known solutions, reliable methods have not been found for the general line-labelling problem, and traditional methods, when adapted to drawings with non-trihedral junctions, are unacceptably slow.Many other papers assume that line labelling is an essential step. Here, we show this is not necessarily true, and that comparable results can be obtained by a novel alternative approach. Firstly, we consider what outputs from line labelling are essential to the production of frontal geometry. Secondly, we investigate by what other means these outputs can be produced.Our work indicates that the only essential output from line labelling for frontal geometry is the determination of which T-junctions in a drawing are occluding and which are non-occluding. This information is required for inflation, and also for detection of symmetry and for constructing hidden topology.Thus, we propose and analyse a new method which, in the absence of line labels, simultaneously inflates a drawing to produce the frontal geometry and attempts to determine whether each T-junction is occluding or not. For drawings of objects with holes or pockets, and for cases where line labelling is particularly unreliable, our new method can provide a better alternative.